profoundSkyPoly: Fit N-order Polynomial Sky

In asgr/ProFound: Photometry Tools

Fit N-order Polynomial Sky

Description

Functions to fit polynomial sky planes to image data. Includes a useful summary plot. These might be preferable in complicated situations where a large amount of the image is masked by source/s, e.g. a large foreground cluster in the centre of the image. In these cases the restrictions of the fits here might interpolated over masked regions more nicely than the default profoundMakeSkyGrid behaviour. The downside is much more care is needed to mask out the image appropriately, hence the inclusion of a quancut argument and diagnostic plots.

Usage

profoundSkyPoly(image, objects = NULL, degree = 1, quancut = NULL, mask = NULL, 
  mode_shift = FALSE, plot = FALSE, ...)
profoundSkyPlane(image, objects = NULL, quancut = NULL, mask = NULL, mode_shift = FALSE, 
  plot = FALSE, ...)

Arguments

image	Numeric matrix; required, the image we want to analyse.
objects	Boolean matrix (1,0); optional, object mask where 1 is object and 0 is sky. If provided, this matrix *must* be the same dimensions as image. If provided then this is used to initially mask pixels for determining the correct sky.
degree	Integer scalar; the polynomial degree to model the sky with. degree = 0 is a pedestal sky, 1 is a linear plane, 2 is quadratic, 3 cubic, 4 quartic and 5 quintic etc. As a convenience setting degree = -1 sets the sky to be 0 throughout (perhaps useful for checking diagnostics).
quancut	Numeric scalar; optional, the level at which to mask out brighter pixels for analysis. This might be useful if objects has not already been computed.
mask	Boolean matrix or integer scalar (1,0); optional, parts of the image to mask out (i.e. ignore). If a matrix is provided, this matrix *must* be the same dimensions as image where 1 means mask out and 0 means use for analysis.
mode_shift	Logical; should the mode be shifted to zero regardless of best polynomical regression found? As long as the non-object, non-mask and non-quancut excluded pixels contain the true sky mode, this is usually a good idea.
plot	Logical; should a diagnostic plot be made for human inspection? Top-left is the original image; top-right the median subtract sky polynomial; bottom-left the final sky subtract image (i.e. image[in] - sky[out]); bottom-right sky pixel density scaled by RMS (green dash is the Normal expectation, which is hopefully 1-1 in the negative domain).
...	Argument to pass to the 'Original' and 'Final' panels of the diagnostic plot when plot = TRUE.

Details

Internally this uses the R lm and poly functions.

profoundSkyPlane is just a convenience function to setting profoundSkyPoly degree = 1 (i.e. a linear plane), since sometimes this confuses people.

Value

sky	Numeric matrix; the predicted sky map that should be subtract from the user supplied sky.
skyRMS	Numeric scalar; the calculated sky RMS. This is only indicative if the input image is from a single detector, in more complex cases a full map as provided by profoundMakeSkyGrid might be required.
lm_out	List; the output of the lm run internally. This might be useful for additional analysis.
good_pix	Integer matrix; mask of the 'good' pix, i.e. those valued 1 were used for analysis, and those valued 0 were not.
skyChiSq	Integer scalar; the Reduced Chi-Square of the sky pixel distribution (the nearer to 1 the better the sky estimation).

Author(s)

Aaron Robotham

See Also

lm, profoundMakeSkyGrid

Examples

## Not run: 
image = Rfits_read_image(system.file("extdata", 'VIKING/mystery_VIKING_Z.fits',
  package="ProFound"))$imDat

profoundSkyPoly(image, degree=0, quancut=0.5, plot=TRUE) #under cut
profoundSkyPoly(image, degree=0, quancut=0.9, plot=TRUE) #about right
profoundSkyPoly(image, degree=0, quancut=0.999, plot=TRUE) #over cut

profoundSkyPoly(image, degree=1, quancut=0.5, plot=TRUE) #under cut
profoundSkyPoly(image, degree=1, quancut=0.9, plot=TRUE) #about right
profoundSkyPoly(image, degree=1, quancut=0.999, plot=TRUE) #over cut

profoundSkyPoly(image, degree=2, quancut=0.5, plot=TRUE) #under cut
profoundSkyPoly(image, degree=2, quancut=0.9, plot=TRUE) #about right
profoundSkyPoly(image, degree=2, quancut=0.999, plot=TRUE) #over cut

profoundSkyPoly(image, degree=3, quancut=0.5, plot=TRUE) #under cut
profoundSkyPoly(image, degree=3, quancut=0.9, plot=TRUE) #about right
profoundSkyPoly(image, degree=3, quancut=0.999, plot=TRUE) #over cut

## End(Not run)

asgr/ProFound documentation built on Feb. 10, 2024, 9:04 p.m.